Abstract
In this paper, we develop a comprehensive framework for measuring and improving API usability, consisting of both a comprehensive measurement framework and an analysis/modeling framework. Our measurement framework consists of both a set of direct measurements of API usability and a set of indirect measurements of factors that may influence API usability. Our analysis/modeling framework validates the measurement framework empirically and establishes predictive relations between its two sets of measurements. Such predictive relations can help us improve API usability through recommended actions applied on its influence factors. Examples from a preliminary case study are included to demonstrate the viability and potential effectiveness of our approach.